Using A Smartwatch To Detect Changes In Brain Health

Brain health, encompassing some cognitive and affectional functions, is 1 of nan awesome nationalist wellness challenges of nan 21st century. According to nan World Health Organization (WHO), much than 1 successful 3 group worldwide unrecorded pinch neurological disorders specified arsenic stroke, epilepsy aliases Parkinson's disease, while much than 1 successful 2 individuals will acquisition a intelligence upset — including depression, worry disorders aliases schizophrenia — astatine immoderate constituent successful their lives. As populations age, these figures proceed to rise.

Even successful patient adults, encephalon wellness fluctuates complete time, reflecting interactions betwixt aggregate factors, including biology influences and individual manner habits. Analysing day-to-day aliases week-to-week changes successful cognitive and affectional functioning is truthful basal to alteration proactive and individualized prevention strategies.

A squad astatine nan University of Geneva (UNIGE) group retired to find whether wearable and mobile technologies could beryllium utilized to show encephalon wellness continuously and non-invasively. To this end, 88 volunteers aged betwixt 45 and 77 were equipped pinch a dedicated smartphone app and a smartwatch. Over a ten-month period, these devices collected "passive" information — without immoderate involution aliases alteration successful participants' regular habits — including bosom rate, beingness activity, slumber patterns, arsenic good arsenic upwind conditions and aerial contamination levels. In total, 21 indicators were analyzed.

Every 3 months, participants besides provided "active" information by completing questionnaires connected their affectional authorities and undergoing cognitive capacity tests.

Emotional states were nan astir accurately predicted by nan artificial intelligence, pinch correction rates ranging mostly betwixt 5% and 10%. Cognitive states, successful contrast, were predicted little precisely, pinch correction rates ranging from 10% to 20%. In different words, AI performs amended astatine forecasting responses to affectional questionnaires than cognitive tests.

Regarding nan relevance of passive indicators, aerial pollution, upwind conditions, regular bosom rate, and slumber variability emerged arsenic nan astir informative factors for cognition. For affectional states, nan astir influential factors were chiefly weather, slumber variability, and bosom complaint during sleep.

This research, supervised by Prof. Katarzyna Wac of nan Research Institute for Statistics and Information Science astatine GSEM and Prof. Matthias Kliegel of nan Cognitive Aging Laboratory astatine nan Faculty of Psychology and Educational Sciences, is portion of nan associated module task Providemus alz. The adjacent shape is already underway. It intends to cod nan aforesaid types of information complete a 24-month period, while examining individual characteristics of participants associated pinch nan highest- and lowest-performing AI models, successful bid to amended understand their applicability successful real-world individualised scenarios.

Source:

Journal reference:

Matias, I., et al. (2026). Digital biomarkers for encephalon health: passive and continuous appraisal from wearable sensors. npj Digital Medicine. DOI: 10.1038/s41746-026-02340-y. https://www.nature.com/articles/s41746-026-02340-y